Salt welds form due to salt thinning by mechanical (e.g. salt-flow) and/or chemical (e.g. salt-dissolution) processes. Despite being ubiquitous in salt-bearing sedimentary basins, where they may trap large volumes of hydrocarbons, little is published on weld thickness and composition. We here use 3D seismic reflection, borehole and biostratigraphic data from the Atwater Valley protraction area of the northern Gulf of Mexico to constrain the thickness and composition of a salt weld. Seismic data image an ‘apparent weld’ at the base of a Plio-Pleistocene minibasin that subsided into allochthonous salt. Borehole data indicate the weld is actually ‘incomplete’, being c. 24 m thick, and containing an upper 5 m-thick halite and a lower 15 m-thick halite, separated by a 4 m-thick mudstone. The age and origin of the intra-weld mudstone is unclear, although we speculate it is either: (i) Late Jurassic, representing material transported upwards from the autochthonous level within a feeder, and subsequently trapped as allochthonous salt thinned and welded; or, perhaps, more likely (ii) Pliocene, representing a piece of salt carapace reworked from the top of and eventually trapped in the now locally welded sheet. We show that 3D seismic reflection data may not resolve weld thickness, with the presence of relatively thin remnant salt lending support to models of welding based on viscous flow. The halite-dominated character of the weld also supports the hypothesis that tectonic purification may occur during salt flow. Supplementary material : uninterpreted seismic reflection profiles, raw biostratigraphic data and a BOEM biostratigraphic chart are available at https://doi.org/10.6084/m9.figshare.c.4183061